What Is Tabata? The 1996 Study Explained

What is Tabata? It’s a specific, brutal style of interval training: 20 seconds of maximum-effort work, 10 seconds of rest, repeated for 8 rounds. Four minutes, start to finish. That’s the whole protocol as it was originally tested, and most of what gets sold as “Tabata” today drifts from it in one way or another.

The name comes from a person, not a Japanese word for exercise. Dr. Izumi Tabata was a researcher at Japan’s National Institute of Fitness and Sports in the mid-1990s when he ran the study that made this interval pattern famous. The training method itself existed before his paper did.

The 20:10 protocol

Work for 20 seconds at an effort you genuinely can’t sustain much longer. Rest 10 seconds. Repeat for 8 rounds. That’s a full Tabata set, and it clocks in at exactly 4 minutes.

The ratio is what defines it (2:1, work to rest), and the round count is fixed at 8. Change either one and you’ve built a different interval workout, which is fine, but it’s not Tabata anymore. A lot of “Tabata classes” quietly stretch the rest to 15 or 20 seconds because 10 seconds isn’t enough recovery to keep moving fast for 8 rounds. That’s a reasonable modification. It’s also a different workout.

Where Tabata comes from

The training method predates the study by years. Irisawa Koichi, who coached Japan’s national speed skating team, used short, near-maximal intervals to prepare skaters for races that demand both a strong aerobic engine and the ability to go all-out repeatedly. Coaches had been running versions of this for a while. What nobody had done was measure it properly in a lab.

Izumi Tabata and his colleagues set out to test whether the protocol delivered what coaches assumed it did, and whether it beat a more conventional cardio approach. That comparison is the whole point of the study.

The 1996 study

The paper is Tabata I, et al. “Effects of moderate-intensity endurance and high-intensity intermittent training on anaerobic capacity and VO2max.” Medicine & Science in Sports & Exercise. 1996;28(10):1327-30.

Seven or eight trained subjects rode cycle ergometers at roughly 170% of VO2max (an intensity you physically cannot hold for more than a handful of seconds) for 20 seconds, then rested 10 seconds, for 7 to 8 rounds per session. They trained like this 4 days a week for 6 weeks. A separate comparison group did something completely different: 60 minutes of steady-state cycling at a moderate 70% of VO2max, 5 days a week, for the same 6 weeks.

By the end, the interval group’s VO2max (a measure of how much oxygen your body can use during hard exercise, and a standard marker of aerobic fitness) went up by about 7 ml/kg/min. That roughly matched the steady-state group’s aerobic gains, despite the interval group training a fraction of the total time. But the interval group also improved anaerobic capacity by 28%. The steady-state group’s anaerobic capacity barely moved.

That’s the finding that made the paper famous: four minutes of near-maximal intervals, four days a week, built both energy systems at once. An hour of moderate cardio, five days a week, only built one of them.

How Tabata trains two systems at once

Most cardio trains your aerobic system, the slower-burning engine that uses oxygen and keeps you going for minutes to hours. Sprinting and heavy lifting mostly train your anaerobic system, the fast, short-burst engine that doesn’t need oxygen and taps out in seconds.

Tabata’s 20:10 pattern sits right at the edge of both. Twenty seconds of true maximal effort is long enough to hammer your aerobic system and short enough to demand the anaerobic system too, and the ten-second rest is too short for full recovery, so by round 4 or 5 you’re running both systems hard just to keep moving. That overlap is why the 1996 study found gains in both categories from one workout. It’s also why the protocol is so unpleasant by round 6.

“Real” Tabata vs. gym-class tabata-style

Here’s the part most fitness sites skip: the study’s subjects were trained young adults working at close to their physiological ceiling, in a lab, on a cycle ergometer, under supervision, measuring their actual VO2max and blood lactate. Real Tabata, as tested, is savage: you finish the last round barely able to keep pedaling.

What most people do in a gym class or an app called “Tabata” (20 seconds of squats, 10 seconds rest, 8 rounds, at maybe 70-80% effort because form and joints matter when you’re using bodyweight moves instead of a stationary bike) is a HIIT circuit borrowing the 20:10 timing. It’s a good, hard workout. It is not the stimulus from the 1996 paper, and it won’t produce a 28% jump in anaerobic capacity in six weeks. That’s not a knock on it. A sustainable version of a brutal lab protocol is usually the more useful one for anyone who isn’t training as a competitive athlete.

Who Tabata is for

Tabata suits people who already have a fitness base and want a short, hard finisher: after a lift, on a day when 45 minutes of cardio isn’t happening, or as a way to test where your conditioning stands. It’s less suited to a first workout back after months off training, to anyone with joint issues that flare under fast repeated loading, or to anyone who hasn’t cleared hard exercise with a doctor.

If that’s you, start with something gentler and build toward it. Our beginner’s guide covers modified ratios that still deliver a real workout. Otherwise, load the 20:10 timer and find out what round 6 actually feels like.

Frequently asked questions

Who invented Tabata training?
Dr. Izumi Tabata didn't invent the protocol himself. He studied it. The 20:10 interval method was designed by speed skating coach Irisawa Koichi for Japan's national speed skating team, and Tabata's 1996 study on it made the protocol famous enough to carry his name.
Is Tabata the same as HIIT?
No. Tabata is one specific, tightly defined HIIT protocol: 20 seconds of near-maximal work, 10 seconds of rest, 8 rounds. HIIT is the broader category that includes Tabata along with dozens of other work-to-rest ratios. Our Tabata vs HIIT guide covers the differences in detail.

Try the timer